Merge with v2.00

Author: niedev

README.md

@@ -65,6 +65,22 @@ The initial download will get the models from GitHub, however <a href="https://w
 If you have a GitHub account and want to be notified when a new release comes out you can do so by clicking, at the top of the page, on "Watch" -> "Custom" -> "Releases" -> "Apply".
 <br /><br />
 
+<h3>RTranslator 3.0 Coming Soon!</h3>
+
+The 3.0 version of RTranslator will be funded through the [NGI Mobifree Fund](https://nlnet.nl/mobifree), a fund established by [NLnet](https://nlnet.nl).
+
+Main changes of this version:
+- The NLLB translation model will be replaced with the option to choose between: Mozilla Bergamot models, Madlad 400 3B, and HY-MT 1.5 1.8B. All these models have better translation quality than NLLB 54B, similar to Google Translate (with HY-MT 1.5 1.8B having much better scores than Google Translate).
+- MLKit will be replaced, making RTranslator 100% open source.
+- Various techniques will be added to improve translation quality, including: beam search, multilingual dictionaries, Tatoeba integration, and more.
+- The app will be released on Play Store and F-Droid.
+- A self-hosted web version of the app for text translation using Mozilla models will be made available.
+
+The first beta of this version will be released between June and August 2026. Stay tuned 🚀
+
+<img src="https://nlnet.nl/logo/banner.svg" width="200px">
+<br />
+
 <h3>What's new in version 2.1</h3>
 
 - **New GUI!** Designed by [Chiara Chindamo](https://www.linkedin.com/in/chiara-chindamo-946053234/).
@@ -186,6 +202,7 @@ If you want to contribute to this project, first of all, thank you 🚀
 If you don't know where to start, go check the [to-do list](https://github.com/niedev/RTranslator/blob/v2.00/TODO_LIST.md), and in any case, before starting, read the [contribution guidelines](https://github.com/niedev/RTranslator/blob/v2.00/CONTRIBUTING.md).
 <br /><br />
 
+
 <h3>Bugs and problems</h3>
 I remind you that the app is still in beta. The bugs found are the following:
 

app/src/main/java/nie/translator/rtranslator/voice_translation/neural_networks/voice/Recorder.java

@@ -26,7 +26,9 @@
 import android.media.AudioManager;
 import android.media.AudioRecord;
 import android.media.MediaRecorder;
+import android.os.Build;
 import android.util.Log;
+import android.widget.Toast;
 
 import androidx.annotation.NonNull;
 import androidx.annotation.Nullable;
@@ -57,7 +59,7 @@ public class Recorder {
     private boolean isManualMode = false;
     public static final int[] SAMPLE_RATE_CANDIDATES = new int[]{16000};
     private static final int CHANNEL = AudioFormat.CHANNEL_IN_MONO;
-    private static final int ENCODING = AudioFormat.ENCODING_PCM_FLOAT;   //original: AudioFormat.ENCODING_PCM_16BIT
+    private static int ENCODING;
     public static final int MAX_AMPLITUDE_THRESHOLD = 15000;
     public static final int DEFAULT_AMPLITUDE_THRESHOLD = 2000; //original: 1500
     public static final int MIN_AMPLITUDE_THRESHOLD = 400;
@@ -75,6 +77,7 @@ public class Recorder {
     private Thread mThread;
     private int mPrevBufferMaxSize;   //the size of the mPrevBuffer (It depends on the settings of the app (prevVoiceDuration))
     private float[] mBuffer;
+    private short[] mBufferShort;
     private int readSize;   //must be smaller than mBuffer.length or the circular mBuffer array will not work
     private int headIndex;
     private int tailIndex;
@@ -107,6 +110,13 @@ public Recorder(Global global, boolean useBluetoothHeadset, @NonNull Callback ca
         mCallback.setRecorder(this);
         this.vad = vad;
 
+        // set the encoding
+        if(Build.MANUFACTURER.equalsIgnoreCase("vivo")){
+            ENCODING = AudioFormat.ENCODING_PCM_16BIT;   //this is to avoid a bug where on Vivo phones the audio recording wouldn't work
+        }else{
+            ENCODING = AudioFormat.ENCODING_PCM_FLOAT;
+        }
+
         // Try to create a new recording session.
         mAudioRecord = createAudioRecord();
         if (mAudioRecord == null) {
@@ -266,6 +276,7 @@ private AudioRecord createAudioRecord() {
             if (audioRecord.getState() == AudioRecord.STATE_INITIALIZED) {
                 readSize = (minSizeInBytes/4)*2;
                 mBuffer = new float[((MAX_SPEECH_LENGTH_MILLIS+1000)/1000)*sampleRate];  //the buffer size will be larger (by one second) than the audio data of duration MAX_SPEECH_LENGTH_MILLIS
+                mBufferShort = new short[((MAX_SPEECH_LENGTH_MILLIS+1000)/1000)*sampleRate];
                 return audioRecord;
             } else {
                 audioRecord.release();
@@ -329,13 +340,13 @@ public void run() {
                     int oldTailIndex = tailIndex;
                     boolean jumped;
                     if (tailIndex + readSize < mBuffer.length) {
-                        size = mAudioRecord.read(mBuffer, tailIndex, readSize, AudioRecord.READ_BLOCKING);
+                        size = readAudio(mBuffer, tailIndex, readSize);
                         tailIndex = tailIndex + size;
                         jumped = false;
                     } else {
-                        size = mAudioRecord.read(mBuffer, tailIndex, mBuffer.length - tailIndex, AudioRecord.READ_BLOCKING);
+                        size = readAudio(mBuffer, tailIndex, mBuffer.length - tailIndex);
                         tailIndex = 0;
-                        int size2 = mAudioRecord.read(mBuffer, tailIndex, readSize - size, AudioRecord.READ_BLOCKING);
+                        int size2 = readAudio(mBuffer, tailIndex, readSize - size);
                         tailIndex = size2;
                         size = size + size2;
                         jumped = true;
@@ -396,6 +407,21 @@ private int getMBufferRangeSize(int begin, int end){
         }
     }
 
+    private int readAudio(float[] audioData, int offset, int size){
+        if(ENCODING == AudioFormat.ENCODING_PCM_FLOAT){
+            return mAudioRecord.read(audioData, offset, size, AudioRecord.READ_BLOCKING);
+        }else{  //ENCODING == AudioFormat.ENCODING_PCM_16BIT
+            int outputSize = mAudioRecord.read(mBufferShort, offset, size, AudioRecord.READ_BLOCKING);
+            // Using the values just read in mBufferShort we convert the values to mBuffer in the ENCODING_PCM_FLOAT format
+            // Tod od this we iterate the section just wrote of mBufferShort, convert each value from ENCODING_PCM_16BIT to ENCODING_PCM_FLOAT and insert these value in the corresponding section of mBuffer.
+            for(int i=offset; i<offset+size; i++){
+                //The range with ENCODING_PCM_16BIT is [-32768, 32767], while with ENCODING_PCM_FLOAT it is [-1, 1], so we convert accordingly
+                mBuffer[i] = (float) mBufferShort[i] / 32768;
+            }
+            return outputSize;
+        }
+    }
+
     private boolean isHearingVoice(byte[] buffer, int size) {   //old method to measure threshold (not used)
         for (int i = 0; i < size - 1; i += 2) {
             // The buffer has LINEAR16 (2 bytes) in little endian.